This article proposes a novel dynamic lumped-parameter thermal model (D-LPTM) to locate and predict the transient/steady-state maximum winding temperatures for both overlapping and nonoverlapping concentrated windings in electrical machines. In addition to conduction thermal modeling, the proposed D-LPTM introduces a compensation thermal resistance to represent the conduction heat transfer capability under distributed loss generation. Both conduction and compensation thermal resistances can be dynamically adjusted to account for the varying hot spot locations, demonstrating high robustness across various load/thermal conditions. Since the proposed D-LPTM is derived completely based on the analytical heat transfer theorem, it can be generally applied to electrical machines with different sizes and geometries. Compared to the finite element (FE) method, the D-LPTM can achieve similar estimation accuracy with significantly less computation effort. Compared to the analytical thermal models (ATMs) that are limited to steady-state temperature predictions, the proposed D-LPTM can predict transient maximum winding temperatures without requiring iterative curve-fitting processes or auxiliary methods, for example, partial FE and ATMs. The proposed model is applied to a prototype electrically excited synchronous machine and validated by both the FE method and experiments.
Dynamic Lumped-Parameter Thermal Modeling for Tracking Maximum Winding Temperature in Electrical Machines
IEEE Transactions on Transportation Electrification ; 11 , 4 ; 8919-8935
2025-08-01
4014420 byte
Article (Journal)
Electronic Resource
English